EFECTOS CLIMÁTICOS Y PRODUCTIVIDAD EN LECHERÍAS DE WISCONSIN: UN ANÁLISIS PRELIMINAR

Transcription

1 EFECTOS CLIMÁTICOS Y PRODUCTIVIDAD EN LECHERÍAS DE WISCONSIN: UN ANÁLISIS PRELIMINAR Lingqiao Qi Graduate Research Assistant, Agric. & Resource Economics (ARE), U of Connecticut (UCONN), USA Boris E. Bravo- Ureta Professor, ARE- UCONN and Adjunct Professor, Ag. Econ., U of Talca, Chile Victor E. Cabrera Associate Professor & Extension Specialist in Dairy Management, and Alfred Toepfer Faculty Fellow, Dairy Science, U of Wisconsin- Madison MAY 26, 2014

2 Objective of the Research General objective: to contribute to the understanding of the effect of climatic variables on dairy farm productivity. SpeciOic objectives: 1. explore alternate deninitions and measures of climatic effects; 2. test alternative stochastic frontier panel data models in the analysis of dairy productivity and climatic effects; 3. perform an empirical analysis using panel data for the state of Wisconsin.

3 Introduction and Motivation The agricultural sector, which remains important to the U.S. economy (USGCRP, 2009), is more sensitive and vulnerable to climate change than other sectors (IPCC, 2014). The livestock sector is particularly vulnerable to hot weather, especially in combination with high humidity, which can lead to signinicant losses in productivity and, in extreme cases, to animal death (Boyles, 2008; Mader, 2003). Climatic conditions also affect feed supplies by innluencing the growth of silage and forage crops (Hill et al., 2004).

4 Introduction and Motivation Dairy industry is the fourth largest agricultural subsector in the United States. There is a signinicant body of animal and dairy science literature, which establishes the susceptibility of dairy cows to extreme weather conditions (Calil et al., 2012; IPCC, 2014). However, the economic literature on this subject remains quite limited.

5 Research Area

6 Contribution Wisconsin is the second largest dairy producing area in the U.S. where winters can be very cold and snowy, and summers hot and humid. Thus, Wisconsin is an ideal location to examine the effects of extreme climatic factors on dairy production. The specinication of our model makes it possible to calculate a total climatic effect as well as partial effects for temperature, precipitation and seasons. This analysis is a novel contribution to the dairy productivity literature.

7 Background: Heat Stress and Cold Stress Heat and cold stress requires cows to increase the amount of energy used to maintain body temperature and less energy is available for milk production (Collier et al., 2011). Heat stress affects feed intake, feed efniciency, milk yield, reproductive efniciency, cow behavior, and disease incidence (Cook et al., 2007; Tucker, Rogers and Shutz, 2007; Rhoads et al., 2009). Cold stress causes animals to consume more feed but to produce less milk, and it also increases milk fat content (Young, 1981).

8 Background: Economic Effect St- Pierre, Cobanov and Schnitkey (2003) calculated the overall effects of heat stress on the U.S. dairy industry at $900 million/yr ($100/cow per year) even when heat abatement systems were in place. The loss would be as high as $1.5 billion/yr ($167/cow per year) without abatement systems. Mukherjee, Bravo- Ureta and de Vries (2013) incorporated an annual average Temperature Heat Index (THI) in a production frontier model and found a signinicant negative effect on output and document cost- effective adaptation. Seo and Mendelsohn (2008) used a discrete choice model to examine how farmers change livestock species and numbers to adapt to climatic change.

9 Background: Measures of Climatic Effects Alternative measures based on temperature and precipitation have been used to incorporate climatic effects in crop and livestock models (e.g., Mendelsohn, Nordhaus and Shaw, 1994; Kelly, Kolstad and Mitchell, 2005; Arriagada, 2005; Schlenker, Hanemann and Fisher 2006; Deschenes and Greenstone, 2007). THI has been developed and widely used (Kadzere et al., 2002) to measure heat stress suffered by dairy cattle. Here we denine winter and summer averages for temperature and precipitation to capture the climatic effect. Much more to do!!!! Using temperature and precipitation directly, instead of an index such as THI, allows for a clear interpretation of the climate effect on the dependent variable of interest.

10 Methodology: General model MILK = f (COW, LAB, FEED, CAP, ANEX, CREX, HOTT, COLT, HOTR, COLR, T, T 2 ) MILK COW LAB FEED CAP ANEX total milk equivalent production in cwt (which is equal to 45.4 kg) of dairy farms per year number of adult cows in dairy farm total hours of labor including family paid and unpaid labor and management, and hired labor 16% protein- mixed dairy feed equivalent in metric tons book value of breeding livestock, machinery and equipment, and buildings, measured in constant 2012 dollars animal expenses including veterinary and medicine, breeding fees, and other livestock expense, measured in constant 2012 dollars CREX crop expenses including chemical, fertilizer, seeds and plants, gas and fuel, rented machinery, and other crop expense, measured in dollars constant 2012 dollars HOTT average temperature (F ) in summer (i.e., June, July and August) COLT average temperature (F ) in winter (i.e., December, January and February) HOTR average precipitation (mm) in summer COLR average precipitation (mm) in winter T time trend T2 time trend square

11 Methodology: Empirical Model Model 1. Pooled SPF model without climatic variables; Model 2. Pooled SPF model with climatic variables; Model 3. True Nixed effects (TFE) model with climatic variables; Model 4. True random effects (TRE) model with climatic variables; Model 5. True random effects model with the Mundlak specinication (TRE- M) and climatic variables.

12 Methodology: Climatic Effects Climatic Effect Index (CEI): is the joint effect of all climatic variables included in the production frontier on output, holding conventional inputs and other variables constant (Hughes et al. 2011) Total CEI: Partial CEI Expressions: Ø CEI for temperature: Ø CEI for precipitation: Ø CEI for summer: Ø CEI for winter:

13 Data Input- output data: Ag. Financial Advisor (AgFA) 958 dairy farms; 52 Wisconsin counties; 17- year period ( ); a total of 9,437 observations. We include 221 farms with information for 10 or more consecutive years, which yields a total of 3,070 observations in 24 counties. A total of 54 farms have data for the full 17- year period. Climate data Parameter- elevation Regressions on Independent Slopes Model (PRISM) maps. Geographic Information System (GIS) techniques are used to calculate monthly mean temperature and precipitation for each county and year.

14 Data: Descriptive Statistics Table 1. Descriptive Statistics for Wisconsin Dairy Farms: (3,070 Observations) Variable Mean Std. Dev. Min Max MILK (cwt=45.4kg) 28,981 39,157 2, ,541 COW (head) ,650 LAB (hour) 6,718 7, ,597 CAP (2012 $) 77,823 97, ,196,189 FEED (metric ton) 699 1, ,488 ANEX (2012 $) 45, , ,188,064 CREX (2012 $) 91,655 95, ,057,084 T HOTT (F) COLT (F) HOTR (mm) COLR (mm)

15 Summary of Results (1) Estimated coefnicients of all conventional inputs are signinicant with the expected positive sign and values (i.e., between 0 and 1). Dairy herd size is the main input in production. Concentrate feed is the second most important input when unobserved heterogeneity is accounted for. But, expenditure on crops is the second most important input when heterogeneity is ignored. This difference suggests that the treatment of heterogeneity in the production frontier deserves attention. The Nive models exhibit decreasing returns to scale ranging from 0.91 (Model 3) to 0.97 (Model 1).

16 Summary of Results (2) Likelihood ratio tests show that climatic variables should be included in the specinication of the production frontier model. The impact of the climatic variables is consistent: * Higher temperature in the summer has a negative effect on output while the opposite is noted in winter; * Higher precipitation has an adverse effect in both summer and winter. Hausman tests for model 4 shows that unobserved heterogeneity is found to be random and correlated with the other regressors. Model 5 which is a TRE with the Mundlak Correction is better.

17 Summary of Results (3) Table 2. Parameter Estimates for Five SPF Models Pooled Models Models Including Unobserved Heterogeneity Variable Model 1 Model 2 Model 3 Model 4 Model 5 W/o Climate With Climate (TFE) (TRE) (TRE- M) lncow *** *** *** *** *** (0.016) (0.016) (0.020) (0.019) (0.020) lnlab *** *** *** *** *** (0.011) (0.011) (0.010) (0.011) (0.010) lnfeed *** *** *** *** *** (0.006) (0.006) (0.008) (0.008) (0.008) lncap *** *** *** *** *** (0.004) (0.004) (0.004) (0.004) (0.004) lnanex *** *** *** *** *** (0.005) (0.005) (0.007) (0.007) (0.007) lncrex *** *** *** *** *** (0.006) (0.006) (0.007) (0.009) (0.007) HOTT *** *** *** (0.002) (0.002) (0.002) (0.002) COLT *** *** *** *** (0.001) (0.001) (0.001) (0.001) HOTR ** *** *** *** (0.0001) (0.0001) (0.0001) (0.0001) COLR * *** ** *** (0.0003) (0.0002) (0.0002) (0.0002) T *** *** *** *** *** (0.003) (0.002) (0.002) (0.002) (0.002) T *** ** *** *** *** (0.0002) (0.0001) (0.0001) (0.0001) (0.0001) Constant *** *** *** *** (0.065) (0.144) (0.173) Level of SigniNicance: ***1%, **5%, *10%

18 Technical EfOiciency (TE) Average TE very similar across models and higher than the results in the meta- analysis by Bravo- Ureta et al. (2007). However, the range of TE values varies considerably. Table 3. Average Annual Technical EfOiciency for Wisconsin Dairy Farms: Year Model 1 Model 2 Model 3 Model 4 Model 5 W/o Climate With Climate (TFE) (TRE) (TRE- M) Average Minimum Maximum

19 Technical EfOiciency (2) Figure 1. Average, Maximum and Minimum Technical EfOiciency for Wisconsin Dairy Farms (Model 1 and 5): TE Model 1 Model 1_Min Model 1_Max Model 5 Model 5_Min Model 5_Max

20 Climatic Conditions (1) Figure 2. Average, Maximum and Minimum Values of Winter and Summer Temperature in Wisconsin: HOTT_Mean (F) HOTT_Min (F) HOTT_Max (F) COLT_ Mean (F) COLT_Min (F) COLT_Max (F)

21 Climatic Conditions (2) Figure 3. Average, Maximum and Minimum Values of Winter and Summer Precipitation in Wisconsin: HOTR_Mean (mm) HOTR_Min (mm) HOTR_Max (mm) COLR_ Mean (mm) COLR_Min (mm) COLR_Max (mm)

22 Climatic Effect The analysis of the climatic effect is key in this paper. According to Model 5, a one- unit increase in temperature (1 F ) in summer leads to a 0.58% reduction in output. In addition, a 1 cm increase in precipitation in summer, leads to a 0.37% reduction in output. Precipitation in winter is also harmful and a 1 cm increase leads to a 0.47% reduction in output. It is interesting to note that a warmer winter has a positive effect and in this case a one- unit increase in temperature leads to a 0.48% rise in output.

23 Climatic Effect Index Temperature has a larger negative impact than precipitation, and the climate effect has a negative effect on production in summer while the effect in winter is positive. The value of total CEI has a small variation between years, but it reveals a slight downward trend over the years. Table 4. Average Annual CEI Values Based on the TRE- M Model Year CEI_total CEI_temp CEI_prep CEI_summer CEI_winter Average

24 CEI and Output Change The data shows wide variability in output with respect the total CEI for the past 17- year period under study AND a slight negative trend indicating that the climate effect has gradually led to declines in output holding all else constant. Figure 4. Annual Output and Total Climatic Effect (CEI) using the RFE- M Model 33,500 33,000 32,500 cwt 32,000 31,500 31,000 30,500 30,000 29, Est. Output Linear (Est. Output)

25 Scenario Analysis In the best case scenario, CEI is denined as: ü The lowest average summer precipitation (53.6 mm); ü the lowest average precipitation in winter (15.2 mm); ü the lowest average temperature in summer (60.3 F ); ü the highest average temperature in winter (30.5 F ). In the worst case scenario, CEI is denined as: ü The highest precipitation in summer (188.3 mm); ü the highest precipitation in winter (77.5 mm); ü the highest temperature in summer (73.2 F ); ü the lowest temperature in winter (9.9 F ). A baseline using the CEI value calculated from 17- year mean for each of the four climatic variables introduced in the model. Table 5. Scenario Analysis CEI Output (cwt) Output Change (%) Baseline ,087 0 Best Case Scenario , % Worst Case Scenario , % Range between the worst and best case scenario is 8,008 cwt.

26 Concluding Remarks This paper introduces four climatic variables into alternative SPF models and derives overall and specinic measures of the climate effect. Climatic effects are signinicant on dairy farming. In particular, higher summer month temperatures are harmful for dairy production, while a warmer winter is beneoicial. Higher precipitation has a negative effect on dairy production in Wisconsin in both seasons. The results also suggest that, ceteris paribus, there is a mild negative association between the climatic effect and dairy farm output over the past 17 years in Wisconsin. Thus, if climate change continues, research and extension efforts will be needed to promote adaptation strategies. FUTURE/ON- GOING WORK (data, Climatic deoinitions, TFP)

27 Thanks! Gracias Lingqiao Qi U. of Connecticut (UCONN), USA Boris E. Bravo- Ureta UCONN and Adjunct Professor, U. of Talca, Chile Victor E. Cabrera U. of Wisconsin- Madison